Skip to contentSkip to conversation

virtio-net: keep vnet header zeroed after processing XDP

PATCH In review Jason Wang 2 participants 3 messages Nov 29, 2018 – Dec 1, 2018

Subsystems (4)
lore
Share & tools
lore

Jason Wang

author
#0 lore raw

PATCH 1/1 virtio-net: keep vnet header zeroed after processing XDP

We copy vnet header unconditionally in page_to_skb() this is wrong since XDP may modify the packet data. So let's keep a zeroed vnet header for not confusing the conversion between vnet header and skb metadata.

In the future, we should able to detect whether or not the packet was modified and keep using the vnet header when packet was not touched.

Fixes f600b6905015 ("virtio_net: Add XDP support")
---
 drivers/net/virtio_net.c | 14 +++++++++-----
 1 file changed, 9 insertions(+), 5 deletions(-)
drivers/net/virtio_net.c +9 -5
index cecfd77c9f3c..ea672145f6a6 100644
@@ -365,7 +365,8 @@ static unsigned int mergeable_ctx_to_truesize(void *mrg_ctx)
 static struct sk_buff *page_to_skb(struct virtnet_info *vi,
 				   struct receive_queue *rq,
 				   struct page *page, unsigned int offset,
-				   unsigned int len, unsigned int truesize)
+				   unsigned int len, unsigned int truesize,
+				   bool hdr_valid)
 {
 	struct sk_buff *skb;
 	struct virtio_net_hdr_mrg_rxbuf *hdr;
@@ -387,7 +388,8 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
 	else
 		hdr_padded_len = sizeof(struct padded_vnet_hdr);
 
-	memcpy(hdr, p, hdr_len);
+	if (hdr_valid)
+		memcpy(hdr, p, hdr_len);
 
 	len -= hdr_len;
 	offset += hdr_padded_len;
@@ -739,7 +741,8 @@ static struct sk_buff *receive_big(struct net_device *dev,
 				   struct virtnet_rq_stats *stats)
 {
 	struct page *page = buf;
-	struct sk_buff *skb = page_to_skb(vi, rq, page, 0, len, PAGE_SIZE);
+	struct sk_buff *skb = page_to_skb(vi, rq, page, 0, len,
+					  PAGE_SIZE, true);
 
 	stats->bytes += len - vi->hdr_len;
 	if (unlikely(!skb))
@@ -842,7 +845,8 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 				rcu_read_unlock();
 				put_page(page);
 				head_skb = page_to_skb(vi, rq, xdp_page,
-						       offset, len, PAGE_SIZE);
+						       offset, len,
+						       PAGE_SIZE, false);
 				return head_skb;
 			}
 			break;
@@ -898,7 +902,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 		goto err_skb;
 	}
 
-	head_skb = page_to_skb(vi, rq, page, offset, len, truesize);
+	head_skb = page_to_skb(vi, rq, page, offset, len, truesize, !xdp_prog);
 	curr_skb = head_skb;
 
 	if (unlikely(!curr_skb))
Signature · 1 line
2.17.1
maintainer
#1 lore raw

From: Jason Wang <[email protected]>
Date: Thu, 29 Nov 2018 13:53:16 +0800

Quoted 11 lines

We copy vnet header unconditionally in page_to_skb() this is wrong since XDP may modify the packet data. So let's keep a zeroed vnet header for not confusing the conversion between vnet header and skb metadata.

In the future, we should able to detect whether or not the packet was modified and keep using the vnet header when packet was not touched.

Fixes f600b6905015 ("virtio_net: Add XDP support")

Applied and queued up for -stable, thanks.

maintainer
#2 lore raw
In reply to Jason Wang

From: Jason Wang <[email protected]>
Date: Thu, 29 Nov 2018 13:53:16 +0800

Quoted 11 lines

We copy vnet header unconditionally in page_to_skb() this is wrong since XDP may modify the packet data. So let's keep a zeroed vnet header for not confusing the conversion between vnet header and skb metadata.

In the future, we should able to detect whether or not the packet was modified and keep using the vnet header when packet was not touched.

Fixes f600b6905015 ("virtio_net: Add XDP support")

Applied and queued up for -stable, thanks.